⚙️ Microsoft caves to the pressure, recalls ‘Recall’

Subscribe | Advertise

Good morning. I hope you all enjoyed a fun, relaxing Father’s Day this weekend.

Geoffrey Hinton said in a recent interview that he thinks AI will soon be far more intelligent than humans. But as cognitive scientist Dr. Abeba Birhane pointed out, his evidence for this is anything but scientific: “His evidence? He ‘believes’ and he ‘is fairly confident.’”

In today’s newsletter: 

AI for Good: A transformation in hearing aids

Nearly 500 million people — around 5% of the world’s population — suffer from disabling hearing loss. The World Health Organization expects 2.5 billion people to suffer some form of hearing loss by 2050. 

Still, studies have found that the bulk of those with hearing loss don’t use hearing aids. One of the reasons behind this reluctance has to do with the impression that the devices just don’t work that well. 

But artificial intelligence is making them better all the time. 

The details: Hearing aids have employed machine learning algorithms for decades. But these algorithms historically have not been powerful enough to tackle the ‘cocktail party’ problem; they weren’t able to isolate a single voice in a loud, crowded room. 

Why it matters: Wang’s work has found that deep learning algorithms, running in real-time, could separate speech from background noises, “significantly” improving intelligibility in hearing-impaired people. 

The tech is beginning to become publicly available, with brands like Phonak and Starkey leveraging deep learning and AI to enhance their hearing aids. 

Microsoft caves to the pressure, recalls ‘Recall’

After facing weeks of mounting pressure over its upcoming “Recall” feature — which was set to launch June 18 — Microsoft announced a soft recall of the program. 

Key points: Microsoft said Friday that Recall will now roll out as a preview available first to members of the Windows Insider Program sometime “in the coming weeks.” 

Microsoft is slowing the release of the feature “to ensure the experience meets our high standards for quality and security.”

The no-background: Microsoft initially announced the feature as a component of its coming line of AI-enabled Copilot+ PCs. The idea of the feature is that it would take constant screenshots of a user’s activity and store them in an AI-searchable database. 

“Thank you to all the privacy and security people who stood up on this one, and the home customers who clearly rejected the feature from the outset,” Kevin Beaumont, one of the researchers who highlighted the risks of Recall said. “I know there’s a whole bunch of people inside MS who think their customers are stupid, but the reality is: We aren’t.”

Together with Brilliant

AI isn’t magic. It’s math.

The best way to really get an edge in the fast-moving world of AI? Master the concepts powering it. 

Brilliant makes it easy — with thousands of bite-size, interactive lessons in AI, programming, math, data and more.

From neural networks to machine learning and beyond, Brilliant takes a first-principles approach to learning. You’ll master core building blocks before moving on to more advanced skills. 

Want to stay ahead of the curve? Join 10 million other proactive learners around the world and start your 30-day free trial today. Plus, readers of The Deep View get a special 20% off a premium annual subscription.

Meta is halting its AI rollout in Europe

The current “AI revolution” is really not much more than a data revolution. For years, the purpose that informed data collection was simple: Targeted advertisements. 

More recently, a secondary purpose has surfaced: The training of AI models. For many users — particularly the artists who showcase their work online — this was a step too far. 

Instagram AI training: Meta said last month that it will use Facebook and Instagram posts to train its AI models. The company said that this is what gives Meta’s AI an “edge” over the competition. 

What happened: After initially greenlighting Meta’s training practices, the Irish Data Protection Commission backpedaled Friday, requesting that Meta “delay” training its LLMs on user content. 

Max Schrems, the chair of NOYB, said that Meta could choose to launch consent-based training. It chooses not to. 

“The Meta press release reads a bit like collective punishment,” he said. “‘If any European insists on his or her rights, the entire continent will not get our shiny new products.’”

💰AI Jobs Board:

 📊 Funding & New Arrivals:

🌎 The Broad View:

*Indicates a sponsored link

Together with Notta Showcase

Go Global With AI Video Translation

In today’s digital era, your audience is global. With Notta Showcase AI video translator, you can effortlessly convert videos into 15+ languages with natural-sounding dubbing, making sure your message reaches & resonates with viewers worldwide.

Unlock the full potential of your videos:

Don't miss out on the opportunity to engage with a global audience. Get started for free today.

Report: GenAI’s explosive growth is ‘cooling’ 

2023 was a year of anxious enthusiasm for generative AI. ChatGPT was novel and no one wanted to get left behind. But the speed of that adoption in the enterprise has, in the face of a less rosy reality, been exchanged for practicality.

Lucidworks’ second annual Generative AI Global Benchmark study found that only 25% of companies have successfully launched AI initiatives over the past year.

The details: The report — which surveyed more than 2,500 business leaders around the world — found that only 63% plan to increase AI spending over the next year. Last year, that number was 93%. 

"The initial wave of enthusiasm for generative AI is being met with a more strategic approach,” Lucidworks CEO Mike Sinoway said in a statement. “Businesses are recognizing the potential of this technology, but they're also cautious about the risks and costs.”

My view: We’re moving into a place where the hype around AI is being tempered by the realities of application. 

It is becoming clear that the realities of applying generative AI are far from magical. Systems — whose biggest pitch is cost-savings — are expensive to deploy. They suffer from unreliability and inaccurate output, since hallucinations are part and parcel with the architecture. And they reveal security vulnerabilities that many corporations can’t choose to ignore. 

Taken together, I think this all adds layers of evidence to the assumption that AI is a bubble about to be introduced to a very sharp needle. 

Good or bad, reality is rarely as exciting as the marketers want you to think it’ll be. This report adds credence to the idea that the future of enterprise AI adoption won’t involve some dramatic transformation; it’ll just involve faster, more powerful apps and software (so long as they can guarantee privacy and add real value).

Have cool resources or tools to share? Submit a tool or reach us by replying to this email (or DM us on Twitter).

*Indicates a sponsored link

SPONSOR THIS NEWSLETTER

The Deep View is currently one of the world’s fastest-growing newsletters, adding thousands of AI enthusiasts a week to our incredible family of over 200,000! Our readers work at top companies like Apple, Meta, OpenAI, Google, Microsoft and many more.

If you want to share your company or product with fellow AI enthusiasts before we’re fully booked, reserve an ad slot here.

One last thing👇

That's a wrap for now! We hope you enjoyed today’s newsletter :)

We appreciate your continued support! We'll catch you in the next edition 👋

-Ian Krietzberg, Editor-in-Chief, The Deep View